He-Ran Wang for Advanced Study at Tsinghua University, Ilya Vilkoviskiy and Dmitry A. Abanin of Princeton University and École Polytechnique Fédérale de Lausanne have identified three qualitatively distinct scalings of temporal entanglement entropy when analyzing quantum circuits. This work establishes a hierarchy of computational resources needed for accurate simulations of quantum many-body dynamics.

The researchers’ analysis reveals that certain quantum dynamics can be efficiently computed using a Monte Carlo algorithm, indicating these systems possess characteristics that allow for this approach. Their results connect quantum many-body dynamics with group theory, offering new insight into the complexity of these systems and their physical characteristics.

Influence Matrix Framework for Quantum Many-Body Dynamics

This hierarchy dictates the computational resources needed for accurate tensor-network representations of the influence matrix (IM), a tool used to characterize quantum many-body dynamics and its multitime correlations. Abanin focuses on a family of brickwork quantum circuits ranging from integrable to chaotic regimes. Beyond these simplified scenarios, the researchers also introduced an operational measure of quantum memory, complete with an experimental protocol, to quantify long-time genuinely quantum correlations.

This diagnostic relies on a quantum teleportation protocol and is designed for cases where a classical IM description is unavailable. The researchers state in their report that the validity of this approach is demonstrated by explicit examples that preserve long-lived quantum memory. The team’s findings establish a new connection between quantum many-body dynamics and group theory, providing fresh insights into the complexity of the IM and its relation to the physical characteristics of the dynamics.

Their work suggests that the notions of classical and quantum memory within the influence matrix offer an alternative dimension for characterizing nonequilibrium dynamics, applicable to generic systems and potentially illuminating the emergence of universal macroscopic phenomena like thermalization and relaxation. The analytical techniques developed are also valuable for exploring solvable many-body systems in broader contexts, including noisy and dissipative dynamics, higher spatial dimensions and cellular automata.

Geometric Group Theory Links to Temporal Entanglement Scaling

The analysis focused on a family of circuits spanning integrable to chaotic regimes, allowing researchers to rigorously connect the scaling of bond dimension, a measure of computational cost, to the growth function, a concept from group theory quantifying how quickly a group can be generated. This connection establishes that the speed of group generation directly relates to the parameterization of local gates within the quantum models.

The team’s work goes beyond simply identifying complexity. It characterizes it through analytical MPS representations of influence matrices, where surprisingly, inner bonds are associated with a group algebra. This allowed for a precise characterization of effective bond dimension scaling and the identification of complexity classes, demonstrating that the computational effort needed to accurately simulate the influence matrix depends on the total number of time steps.

A key finding is the identification of models where their multitime correlation functions can be replicated by classical stochastic processes. This enables efficient Monte Carlo simulations for local dynamics, a surprising result given the typically intractable nature of quantum systems with classical methods.

The researchers uncovered a subset of models where the influence matrix admits an efficient Monte Carlo algorithm for computing multitime correlations, suggesting these systems behave as if governed by classical dynamics. The paper states that these models are “abelian up to a finite part,” hinting at the simplified structure enabling this classical approximation.

Polynomial Complexity in Non-Interacting Quantum Environments

Analytically establishing the computational complexity of the influence matrix (IM) has remained limited to a small number of examples, yet recent work expands understanding of these complexities in non-interacting fermionic environments where polynomial scaling of IM complexity has been rigorously established. Previous studies demonstrated similar results for free bosons, and area-law entangled solutions for the “Rule 54” cellular automaton, but a unified framework for exploring the complexity of nonequilibrium quantum dynamics remained elusive.

Researchers have now analyzed a family of brickwork quantum circuits to address this gap, providing a flexible setting for investigating the intricacies of these systems. Further analysis revealed linear TEE growth for generic circuit realizations within their model, establishing three distinct classes of computational complexity for simulating quantum dynamics.

Analysis of quantum and classical memory further refined this understanding, distinguishing between the types of temporal correlations present in different models. The paper states, “We now turn to another notion of temporal complexity, defined in terms of classical and quantum memory,” outlining the next step in characterizing these complex systems.

Integrable Models Exhibit Sublinear Temporal Entanglement Entropy

Analysis of quantum circuits reveals a tiered structure to computational complexity, with three distinct ways temporal entanglement entropy scales with system size. The team identified that certain models exhibit sublinear growth of temporal entanglement entropy, mirroring results previously seen in free bosons and area-law entangled systems. However, this analysis goes further, establishing a relationship between the computational demands and the underlying physics of the system.

Specifically, the scaling of bond dimension within matrix product state representations dictates computational class, with polynomial scaling indicating sublinear temporal entanglement entropy growth. Numerical results suggest this temporal entanglement grows slower than conventional spatial entanglement, even in systems demanding exponential computational resources. A surprising discovery was the existence of specific matrices within certain models. This suggests that, for these specific quantum dynamics, classical algorithms can accurately model the system’s evolution.

Dual-Unitary Circuits and Thermalizing Initial States

Analysis of influence matrices (IMs) reveals distinct scaling behaviors of temporal entanglement entropy (TEE) within brickwork quantum circuits, establishing a tiered hierarchy of computational demands for accurate tensor-network representations. Previous conjectures posited sublinear TEE scaling in integrable models. The researchers extended this understanding by studying a novel family of brickwork quantum circuits, allowing exploration of the complexity of nonequilibrium quantum dynamics beyond simple product-state IMs.

Their approach uses single-site unitaries paired with swap gates, constructing a folded representation where forward and backward time evolution branches are combined. This construction applies to generic matrix product state (MPS) initial states, going beyond previous limitations. For solvable initial states, the IM becomes perfectly Markovian, a finding validated through exact matrix product operator (MPO) constructions.

This establishes regimes of no memory, classical memory and genuine quantum memory, forming a complementary hierarchy of temporal complexities. The researchers identified a subset of their models that have classical influence matrices, in the sense that the multitime correlation functions can be efficiently computed, suggesting these systems possess effectively classical IMs.

Brickwork Quantum Circuits Define Complexity Hierarchy

This goes beyond simply acknowledging simulation complexity. It details how that complexity scales, revealing different demands for different quantum dynamics. The work builds on previous observations of area-law temporal entanglement entropy in dual-unitarity setups, but moves toward a more unified understanding of the connection between IM complexity and the underlying physical characteristics of the dynamics. The researchers demonstrated that certain realizations of their model exhibit persistent quantum memory at long times, opening an additional dimension of temporal complexity for characterizing IMs and nonequilibrium quantum dynamics.

In contrast, generic circuit realizations within the model displayed linear temporal entanglement entropy growth, suggesting a fundamentally different computational cost for simulating their dynamics. These findings suggest that the previously fragmented picture of connections between IM complexity and dynamical properties is beginning to coalesce.

Growth Function Determines Bond Dimension Scaling

The computational demands of simulating quantum many-body dynamics are directly tied to a mathematical concept known as the growth function, according to new analysis of brickwork quantum circuits. Researchers have established a hierarchy of computational complexity based on how quickly a group, a set of mathematical symmetries, expands from its initial elements, linking this expansion rate to the necessary bond dimension for accurate tensor-network representations of quantum systems.

This connection provides a rigorous way to characterize the scaling behavior of these simulations, moving beyond simply acknowledging their inherent difficulty. The team discovered that the size of the group elements at a given time step directly determines the effective bond dimension needed to represent the system’s influence matrix (IM). This reachable set, denoted as H(T), grows over time, and its asymptotic growth rate, the growth function, dictates the computational cost of accurately simulating the IM.

A slower growth function translates to a lower bond dimension, and therefore a less computationally intensive simulation. The researchers derived an exact matrix product operator representation of the IM, associating inner bonds with a group algebra. This allows them to rigorously connect bond dimension scaling to the growth function, establishing a tiered hierarchy of computational resources.

The work demonstrates that the temporal scaling of bond dimension is not arbitrary, but is instead governed by the underlying algebraic structure of the system’s dynamics. This means that understanding the group-theoretic properties of a quantum system can predict the resources required for its simulation, offering a pathway to optimize simulation strategies.

Quantum Memory Measure Captures Genuine Correlations

This efficiency arises in cases where the IM, despite originating from a fully quantum circuit, behaves as a purely classical memory structure, allowing for a classical approach to calculating multitime correlations. Researchers detailed this finding while investigating the interplay between classical and quantum temporal correlations within the IM, opening a path to tackle previously inaccessible regimes with tensor-network methods. This measure addresses scenarios where the IM lacks a classical description, allowing for the assessment of long-time genuinely quantum correlations.

The approach moves beyond simply acknowledging the presence of quantum effects, instead providing a means to characterize the amount of quantum information retained within the system’s dynamics. This new measure is validated through numerical realization within their model, offering a concrete tool for future investigations. The researchers demonstrated that for certain initial states, the IM of their model is indeed classical, enabling simulation via Monte Carlo sampling algorithms.

👉 More information
🗞 Temporal complexity hierarchies in solvable quantum many-body dynamics
✍️ He-Ran Wang, Ilya Vilkoviskiy and Dmitry A. Abanin
🧠 DOI: http://link.aps.org/doi/10.1103/3v7h-7dx9

More like this